Does sampling using random digit dialling really cost more than sampling from telephone directories: debunking the...

Area of Science:

• Public Health
• Survey Methodology
• Health Services Research

Background:

• Computer assisted telephone interviewing (CATI) is a common method for health surveys, offering timeliness and cost benefits over face-to-face interviews.
• Two primary CATI sampling methods include electronic white pages (EWP) and list assisted random digit dialling (LA-RDD).
• EWP sampling uses listed household numbers, while LA-RDD offers broader coverage but is perceived as more costly due to out-of-scope numbers.

Purpose of the Study:

• To compare EWP and LA-RDD sampling procedures in health surveys.
• To evaluate demographic representation, health estimates, and cost-effectiveness of each sampling method.
• To determine which method achieves a similar level of estimate precision more efficiently.

Main Methods:

• A comparative analysis of EWP and LA-RDD samples from the 2003 New South Wales Population Health Survey.
• Evaluation metrics included demographic profiles, health estimates, coefficients of variation in weights, and design effects.
• Cost-effectiveness was assessed based on achieving equivalent precision in health estimates.

Main Results:

• LA-RDD samples demonstrated superior population representation compared to EWP samples (CV of weights: 1.03 vs 1.21).
• LA-RDD exhibited lower average design effects (2.00 vs 2.38), indicating greater precision.
• LA-RDD sampling achieved up to 14.2% cost savings for equivalent precision in health estimates.

Conclusions:

• LA-RDD sampling provides better population representation, potentially reducing bias in health survey estimates.
• Contrary to expectations, LA-RDD sampling is more cost-effective than EWP sampling for achieving desired precision.
• LA-RDD is recommended for health surveys requiring accurate and efficient data collection.